Hamer 



Chapter 17 



Inland Habitat Associations in Western Washington 



blooms help trap debris falling from the upper canopy, creating 

 additional nesting platforms and platforms of larger size. 



Describing Low and High Quality Habitat 



In order to use the model to predict the probability of 

 occupancy of an old-growth stand by murrelets, and thus 

 judge the suitability of a stand as nesting habitat, it is necessary 

 to obtain values for the 8 variables used by the model from 

 the stand needing evaluation. The values for these variables 

 can then be compared to the mean, minimum, and maximum 

 values calculated for stands with a high probability of 

 occupancy and stands with a low probability of occupancy 

 (table 4). Using this comparison, a general sense of the 

 suitability of a stand as nesting habitat can be obtained. In 

 addition, by entering the values for the 8 forest characteristics 

 into the formula shown below, the probability of occupancy 

 can be calculated. Elevation should be entered in feet, stem 

 density as the number of trees/25 m plot, mean d.b.h. of 

 western hemlock in cm; and lichen, moss and canopy cover 

 as percent total cover. First, the logistic regression model is 

 used to predict g(x) as follows: 



g(x) = 

 where. 



(1) 



b is the intercept and, b v ...., b g are the logistic regression 

 coefficients for each variable. These values are listed under 

 the Regression Coefficients in table 3. 



.v, , ...., Jt g are the values for the independent variables 

 measured at the stand in question and, 



g(x) is the predicted value of the logistic transformed 

 probability of occupancy. 



Then g(x) is retransformed to estimate the probability of 

 occupancy as follows: 



P = EXP (g(x))/( 1 + EXP (g(x))] where, (2) 

 P is the predicted probability of occupancy, 

 g(x) is as defined in equation (1). 

 EXP is the exponentiation function, i.e. 



EXP 3 = e 3 where e = 2.7183..., the base of natural 

 logarithms. 



It is important to recognize that this model was developed 

 from a sample of old-growth stands and its reliability in 

 other stands has not been evaluated. 



Tree Characteristics 



Because western red cedar ranked third in producing 

 potential nest platforms and was indicated by the regression 

 analysis to be helpful in assessing suitable habitat, nest- 

 search parties should pay closer attention to this conifer. 



Western hemlock was rated lower as a suitable nest tree 

 because of a lower platform abundance. Because observers 

 did not count mistletoe brooms on the outer limbs of trees as 

 potential nest platforms, the actual number of potential nest 

 platforms/tree for western hemlock may be much higher. 



Because murrelet surveys are often conducted in stands 

 containing a mix of conifer species, it is difficult to use 

 detection trend information from different stand types to confirm 

 a preference for nesting in one type of conifer. In addition, not 

 enough murrelet nests have been located, or located in a 

 random manner, to determine whether birds are selecting 

 particular tree species for nesting, especially since greater 

 nest-search and survey effort have occurred in the Douglas-fir 

 and Western Hemlock zones than in the Sitka Spruce zone. 

 This comparison provides evidence that certain tree species 

 are more likely to be used by murrelets than others. 



Acknowledgments 



These studies were funded by the Washington Department 

 of Wildlife Nongame Program and Pacific Northwest Region 

 Office, USDA Forest Service. I am grateful to Eric Cummins 

 and Bill Ritchie of Washington Department of Fish and 

 Wildlife for their considerable contributions of time and 

 energy in developing and carrying out this research. Additional 

 funding and field personnel were obtained from the 

 Washington Department of Natural Resources Forest Land 

 Management Division. The Endangered Species and 

 Migratory Bird Programs of the U.S. Fish and Wildlife 

 Service, U.S. Department of Interior, in Portland, Oregon, 

 also contributed valuable funding. Dick Holthausen and Grant 

 Gunderson (USDA Forest Service) were both instrumental 

 in coordinating the participation of the Forest Service in the 

 project. I thank Lenny Young and Chuck Turley of the 

 Washington Department of Natural Resources for their support 

 of the research and Tara Zimmerman of the U.S. Fish and 

 Wildlife Service for her efforts in providing additional funding. 

 I thank Phyllis Reed and Charlie Vandemoer of the USDA. 

 Forest Service, Mt. Baker-Snoqualmie National Forest for 

 their cooperation and help with logistical needs. 



I also acknowledge the large number of field personnel 

 from the Washington Department of Fish and Wildlife, Mt 

 Baker-Snoqualmie, Olympic, and Gifford Pinchot National 

 Forests, Washington Department of Natural Resources, private 

 biological consulting companies, and the timber industry for 

 their major contributions to data collection and willingness 

 to share information. I thank statisticians Tim Max and Don 

 Bachman of the USDA Forest Service Forestry Sciences 

 Laboratory biometrics group for their help in study design, 

 analysis, and interpretation of a complex habitat model. 



Helpful reviews of this manuscript were provided by 

 Alan Burger, Martin Raphael, C. John Ralph, Peter Conners, 

 Dean Stauffer, Bill Block, and Jim Baldwin. 



USDA Forest Service Gen. Tech. Rep. PSW-152. 1995. 



175 



